Projects per year
Due to the scarcity of water resources and stricter government regulations, water recycling in the mining industry is becoming a key solution to save water and build zero-emission concentrators. However, such action can have dramatic impacts on the performance and maintenance of the concentrators due to variations in process water quality. This study reveals the convenience of (1) multivariate data analysis for evaluation and interpretation of large water quality datasets and (2) multivariate statistics for monitoring water quality. The aims are to acquire a better understanding about historical temporal variation of water quality due to seasonal variation and/or water circuit modifications and to introduce a multivariate statistics method for monitoring process water quality in the mining industry. The data matrix (797 observations) was treated with Principal Component Analysis (PCA), to extract the variability and to detect the major changes when the processing plant transitioned from a long water cycle to a short water cycle. Additionally, multivariate statistics parameters such as the Q residual and Hotelling's T2 were used for detecting shift in water quality and associated causes. This method is known as Multivariate Statistical Process Control (MSPC). The Q residual provides information about the correlations between variables and Hotelling's T2 gives information on operating ranges of the inputs. Compared to univariate control, MSPC shows several advantages: (1) reducing the number of monitor charts needed, (2) increasing the signal to noise ratio and (3) taking into consideration all parameters and their correlation. Because of these advantages, multivariate analysis and MSPC can be an extremely useful tool for mineral engineers and operators to control the water quality in the plant and to make decisions on process/water circuit modifications. With MSPC, the water quality can be maintained in a controlled range and the metallurgists only need to deal with feed variations to maintain the plant performance.
- Multivariate data analysis
- Multivariate Statistical Process Control (MSPC)
- Water quality monitoring
- Water recycling